{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": [],
    "toc": true
   },
   "source": [
    "# WS 15 Schwellenwert für Accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* `.predict_proba()` gibt bei allen Klassifikatoren die Wahrscheinlichkeit für die Zugehörigkeit zu den einlenen Klassen zurück, `predict()` dagegen die wahrscheinlichtste Klasse selber\n",
    "* für Zwei-Klassen Fragestellungen bedeutet dies, dass bei einer Wahrscheinlickkeit (proba) `> 0.5` für die erste Klasse diese zurückgegeben wird, andernfalls die zweite Klasse, `0.5` ist somit ein scheinbar willkürlicher Schwellenwert\n",
    "* untersuchen Sie die Auswirkung anderer Schwellenwerte auf die Accuracy mit `RandomForestClassifier` auf den aufbereiteten Bankkunden-Datan\n",
    "\n",
    "* vorgeschlagenes Vorgehen:\n",
    "  * trainieren eines RandomForestClassifier mit den vorbereiteten Bankkundendaten (Trainingsdaten)\n",
    "  * bestimmen der Wahrscheinlichkeit für jede Beobachtung der entsprechenden Testdaten zur Klasse 'no'\n",
    "  * erstellen einens Range der zu untersuchenden Schwellenwerte, z.B. mit np.arange()\n",
    "  * in einem Loop über alle Werte dieses Ranges\n",
    "    * `y_pred` für den jeweiligen Schwellenwert berechnen (wiederum als `['no', 'yes']`)\n",
    "    * `accuracy_score()` der jeweiligen Prediction (und sammeln in einer Liste)\n",
    "  * ausgeben des besten Score-Wertes und des zugehörigen Schwellenwertes in der Konsole\n",
    "  * visualisieren der Ergebnisse auch als Lineplot  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "## prepare env, read and prepare data\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns; sns.set()\n",
    "\n",
    "codepath = '../2_code' ## for import of user defined module\n",
    "datapath = '../3_data'\n",
    "\n",
    "from sys import path; path.insert(1, codepath)\n",
    "from os import chdir; chdir(datapath)\n",
    "\n",
    "from bfh_cas_pml import prep_data\n",
    "X_train, X_test, y_train, y_test = prep_data('bank_data_prep.csv', target='y', seed=1234)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-04-15T21:03:01.918701Z",
     "start_time": "2020-04-15T21:03:01.844142Z"
    }
   },
   "outputs": [],
   "source": [
    "## train a model\n",
    "from sklearn.ensemble import RandomForestClassifier \n",
    "model = RandomForestClassifier(random_state=1234)\n",
    "model.fit(X_train, y_train) \n",
    "\n",
    "## prediction using .predict_proba()\n",
    "y_pred_p_no = model.predict_proba(X_test)[:, 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0\n",
      "0.1\n",
      "0.2\n",
      "0.30000000000000004\n",
      "0.4\n",
      "0.5\n",
      "0.6000000000000001\n",
      "0.7000000000000001\n",
      "0.8\n",
      "0.9\n",
      "1.0\n"
     ]
    }
   ],
   "source": [
    "## inspect different threshold values\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "thresholds = np.arange(0, 1.01, 0.1) ## test over 10\n",
    "#thresholds = np.arange(0, 1.01, 0.01)\n",
    "\n",
    "scores = []\n",
    "\n",
    "for threshold in thresholds:\n",
    "    ## tbd\n",
    "    print(threshold)\n",
    "\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "## results\n",
    "## tbd\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "## viszalization\n",
    "## tbd\n",
    "\n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.7"
  },
  "toc": {
   "base_numbering": "",
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "WS 15 Validierung - Sampling",
   "title_sidebar": "Contents",
   "toc_cell": true,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "195.867px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "position": {
    "height": "306.85px",
    "left": "862px",
    "right": "20px",
    "top": "137px",
    "width": "350px"
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}